Mavic2 jerky waypoint issue solved at last!
The firmware update just released by DJI fixed their issue with jerky waypoints. This was a problem with any flight automation software.
The quick footage I just took with Dronelink to confirm this showed the defect is finally put to bed.
By using the "Curved" interpolation settings easing/ramping transitions in and out I was able to get some extremely smooth footage.
Can't wait to have more time to put DL and the M2P through their paces now that DJI has finally corrected the defective WPs.
I don't think flight automation in a portable solution can get much better than a M2P running Dronelink!
Comments
44 comments
Awesome!
glad to hear this! I just ran a M2P mission today and was disappointed by this behavior.
So you haven't updated your firmware yet for the M2P? If not, don't forget to update all your batteries and RC after the M2.
I updated the FW and tested this as well. As you say - the issue is now fixed! Dronelink needs a good signal between RC and Drone though - in order to fly smoothly. If you push it too far one can experience some jerkiness in rare cases.
I still notice a little jerkiness when flying slowly after the fw upgrade (especially when changing direction). Anyone else see this on the M2P?
My flights have so far been relatively fast in somewhat windy and gusty conditions. It was obvious that things were much improved. When the weather improves I will do some slow speed runs.
Great news, guys.
Yes - I do see this to some degree on M2P after FW fix. Not much, but not as smooth as native waypoint flight.
I ran a test this morning and still see jerky panning movements. I agree with Hakon Lislebo, this is not as smooth as a native waypoint flight since DJI fixed the firmware.
Jim McAndrew I just completed a reasonably complex test flight this morning and everything went beautifully. None of the previous errors I had encountered. I love being able to set up the camera in the mission so that I never forget or risk a mistake on site. And being able to do everything (path, orbit, and map) all in one mission is fantastic.
Now that I have taken the time to go through the tutorials, this jerky panning could be the only thing stopping this app from becoming my #1 tool. But, that is a big one. Is it something you can take care of? Do you need to do something with an updated SDK perhaps to take advantage of the new firmware?
Thanks!
The jerky yaw behavior that was being caused by DJI's firmware was a separate issue from what you are now seeing. Dronelink runs on the mobile device and has to overcome a 100ms latency to achieve smooth results. The current options are either slow the mission down or run it on a larger airframe (such as a P4 or Inspire). We may invest more time in to optimizing the flight controller for smooth, cinematic shots at high speeds on the M2, but we are currently focused on more commercial and industrial inspection use cases for our enterprise customers where this is not an issue.
In the meantime, I was able to take it out on my shots using the stabilizer in davinci resolve (free).
Jim McAndrew I just tried it with my P4P. The majority of it was much smoother. There was one bit right at the start where it was actually worse, but that was a super fast turn.
Waa Productions I have Davinci Studio - I'll give that a shot - it can often salvage these things. Thanks
Jim McAndrew, I came here to see if there was an explanation of a 12+ minute mission I ran that consisted of mostly two orbits. The majority of the mission was smooth, but there were definitely moments of slightly jerky yaw.
I have a M2Z (updated firmware to fix the jerky issue)), standard controller, and a Galaxy Tab A w/S-Pen (2016) tablet. It runs the DJI4Go software perfectly, as well as Litchi and Dronelink.
Because the mission consisted of two large orbits, the orbit speed was 14mph. It sounds like the footage was a victim of what you described in this thread - latency while moving at moderate speed. I had line of sight the entire time, with the drone at 250ft the entire mission.
I will try the flight again (not sure when I'll be at the location again) and reduce the speed to 8mph (and probably break up the mission into two missions with one orbit each due to time-in-flight).
Is 8mph reasonable? Should 14mph be reasonable?
In all, the mission planning and mission execution was awesome.
Thanks...
It's less about the forward speed and more about the required angular velocity to maintain the desired heading. Slowing down the plan will reduce the required angular velocity and help it out.
Thanks, Jim.
For reference, my mission had some waypoints and 2 relatively big orbits. The first orbit had an 838ft radius, drone heading 90 degrees, 14mph, and a gimbal angle of -10. The second orbit had a 1,118ft radius, heading 270 degrees (looping counterclockwise), 14mph, and a gimbal angle of -13.
I may try breaking this up into two missions with 1 orbit each and slowing down to 8mph. The single mission at 8mph would run about 20 minutes.
We have a potential fix for this issue on our dev servers but testing it is time consuming. Let me know if you would like to help test before it goes to production.
Thank's really good news! Thanks to you and the dev team for tackling this!
Unfortunately, the weather in the Northeast is pretty darn cold and I won't have an opportunity to try this out for a while (thanks for the offer!)
Is the update in the app or online?
Hi JIm.
I could take a look at the weekend if that works.
Thanks
Barry
That would be great. Do you still have access to the dev server?
I don't know. I can check tonight. Can you send the link to save me searching for it.
I can test during the weekend if the weather allows.
I ran a couple of test missions today with the hope of seeing an improved pan around objects using the M2P.
In fairness, I made it deliberately difficult. Speeds were pretty fast and the distances were not large.
The two missions were as follows:
https://app.dev.dronelink.com/barry-houldsworth/barry/plan/swdduVBLLGODQGgPMDO2/zmUcFrqSTOu0QGe5DbPN
https://app.dev.dronelink.com/barry-houldsworth/barry/plan/JdCNtog7hxNByvcka1qu/wA6hbHUCnfZgzzUsyStS
I was using an iPad Pro 9.7in and Dronlink for DJI - Dev Version 1.1.1 (51)
In the first mission, I set a single point of interest on my house but, from what I could see, the house didn't stay in the center of the frame and the movements around it were not as smooth as I would expect.
I ran this mission twice and the second one seems noticeably better than the first one - much smoother.
Here is flight #1: https://youtu.be/6fIqkbR48AM
Here is flight #2: https://youtu.be/fQKnetox5wI
I used the second mission because I had run that before and had some seriously jerky results. That one was actually much improved over the last time I ran it and would probably consider that acceptable.
I also had an orbit in the second mission that at the beginning struggled to get going and I actually had the mission abort for being too far off course. I restarted and the orbit from that point on was actually well within the bounds of acceptable.
One thing to mention is that as soon as I took off I got a message about "Electromagnetic interference" and that the remote controller may fail to provide control. But...I don't believe that. I used DJI Go 4 just before and after and received no such message. I've also had this before, but only with the M2P and only using DroneLink, so I feel like it's a bogus message.
I hope that helps!
Thanks Barry!
The EM interference and obstacle warnings are both DJI UX SDK issues that I am hoping they will correct in the next release.
As for the mission plans above, I ran the first once and also saw it struggling to keep center, and from time to time it would speed up the yaw in an attempt to try to catch up. On the off-center issue, in my case it seemed to be due to this issue:
https://support.dronelink.com/hc/en-us/community/posts/360034749773-Seeming-to-have-a-Heading-error-
I am working on another potential fix to get around this (which I believe is a firmware issue), but for the purpose of the bug that this thread is about, I think the behavior is no worse than before.
On the increasing speed to catch up issue, I was carefully watching the expanded mission view and looking at what yaw commands were being sent to the drone, and in my case it was constantly around 7 degrees/second during those portions of the mission, which leads me to believe that it was actually gimbal movement causing the apparent motion, and I am guessing it is related to the other issue linked above.
The original issue that I believe this thread was really about was small, intermittent jumps that the drone and/or gimbal would take when rotating at relatively tame speeds (less than 2 degrees / sec) and when focusing on relatively distant POIs, and my fix was definitely focused on those cases. I suspect that there may need to be a different fix for the issues you and I saw, and it may be related to the other fix that I am already working on. My question is, is the type of motion in your first mission plan something you would actually use on a real mission, or was this mission just designed to push the controller to the limit?
Thanks Jim.
Rotating around a home is a shot that I would often capture when making realtor videos. But I would never do it at those speeds. I'm usually closer to 8-10mph and it's possible that at those speeds it would be much improved. So, to answer your question, I was trying to push the controller.
One more comment. I just read the thread you linked to and there may well be something to that. When the Mavic landed I thought that the gimbal looked a little off to one side. But then after landing it seemed to correct itself and I assumed it was my imagination.
It is not your imagination. You can tell this is happening during a mission by looking at the compass widget (see the green gimbal shade is not aligned with the front of the drone). Of course after it lands you can see it directly, but I have seen the gimbal "settle" back to a neutral position after landing.
If I get a chance to run it again I can keep an eye on the compass widget next time. But, at the moment, it looks as though that won't be for a while due to other commitments.
One thought - would setting the gimbal mode to FPV eliminate or reduce the yaw to stop that happening? That obviously wouldn't produce useful footage, but might help to confirm that the problem is a gimbal problem and not a yaw problem.
Since the office was dead I made it home in time to run a quick follow up test. I tried two variants.
1. FPV mode.
I added a component before the main path that (supposedly) switched it into FPV mode. I'll be honest - I don't think that it changed anything. Have you ever tested that works? When it landed I switched to Go 4 (since I don't think I can see the FPV/Follow mode in DroneLink) and it still showed as follow. So I don't know we learned anything here but it didn't seem to change anything.
2. Slower
I dropped the speed to 10mph instead of 25mph. It was still off-center but the turns were smoother to the point that it is probably useable, at least as long as the gimbal doesn't drift too much.
When it landed I confirmed that the camera was noticeably pointing off to the right. Not my imagination.
I just retested the gimbal model (FPV) setting and it worked for me.
Great. Probably something I did wrong then. Did you do that in conjunction with a curve to see if it stayed on track and didn't drift?
Please sign in to leave a comment.